RFID multi-tag anti-collision identification method for high-density livestock and poultry breeding environment

Abstract

The application discloses a RFID multi-tag anti-collision identification method for high-density livestock and poultry breeding environment, comprising the following steps: calculating the total power level number according to the power scanning parameter and constructing a power gradient set; controlling the reader to emit multi-level power to the livestock and poultry tag group according to the power gradient; under the emission power of each level, activating the livestock and poultry tags in the current power level coverage range, forming a subset of tags to be identified and initiating an identification frame in the current power level, and counting the number of idle time slots, the number of successful time slots and the number of collision time slots; estimating the number of remaining tags in the current power level based on the number of successful time slots and the number of collision time slots; calculating the initial frame length, adaptively correcting the parameters of the next identification frame based on the fuzzy adjustment mechanism; and continuing to identify the remaining livestock and poultry tags in the current power level according to the corrected frame length until all power levels are identified. The application can reduce the channel collision probability in the tag identification process.

Description

Technical Field

[0001] This invention relates to the fields of radio frequency identification and information technology in livestock and poultry farming, specifically to an RFID multi-tag anti-collision identification method for high-density livestock and poultry farming environments. Background Technology

[0002] With the development of smart farming and refined management, RFID technology has been widely applied to livestock and poultry identification and end-to-end traceability. In large-scale farming and slaughtering, livestock and poultry typically exist in high-density groups, leading to a sudden surge of RFID tags within the reader's coverage area. However, the livestock and poultry farming environment has significant unstructured characteristics. First, livestock and poultry, as living targets, exhibit irregular movements, and biological tissues absorb and attenuate electromagnetic waves; second, the metal structures of the livestock pens generate strong multipath reflections. These factors cause drastic fluctuations in signal strength within the channel, easily triggering the near-far effect and the capture effect, where tags with closer proximity or stronger signals continuously suppress weaker signals, creating blind spots in identification.

[0003] Existing RFID anti-collision algorithms, such as dynamic frame time slots, are mostly designed based on ideal static scenarios, lacking physical layer diversion mechanisms, and their parameter adjustment strategies are relatively lagging. When faced with the drastic changes in channel conditions caused by high-density crowding and rapid movement of livestock and poultry, existing algorithms often struggle to achieve real-time and accurate matching of frame length parameters, resulting in severe channel congestion, low inventory efficiency, and an inability to meet the stringent requirements of automated production lines for real-time and complete data acquisition. Summary of the Invention

[0004] In order to solve the above-mentioned technical problems, this application proposes the following technical solution:

[0005] This application provides an RFID multi-tag anti-collision identification method for high-density livestock and poultry farming environments, including:

[0006] The total number of power levels is calculated and a power gradient set is constructed based on the set power scan parameters.

[0007] The RFID reader is controlled to transmit power to a high-density distribution of livestock and poultry RFID tags in a multi-level power gradient manner according to the power gradient set.

[0008] At each level of transmission power, activate the livestock and poultry tags that are within the coverage area of ​​the current power level and have not been set to a silent state, form a subset of tags to be identified at the current power level, and initiate an identification frame within the current power level. Count the number of idle time slots, successful time slots and collision time slots in the current identification frame to form a time slot state vector.

[0009] The number of remaining tags in the current power layer is estimated based on the number of successful time slots and the number of collision time slots;

[0010] The initial frame length is calculated based on the estimated number of remaining tags, and the parameters of the next identification frame are adaptively corrected based on the fuzzy adjustment mechanism.

[0011] Based on the corrected frame length, continue to identify the remaining livestock and poultry tags in the current power level until all tags in the current level are successfully identified and set to a silent state. Then switch to the next power level to continue identification until all power levels have been identified.

[0012] In one possible implementation, the formulas for calculating the total number of power levels and constructing the power gradient set based on the set power scanning parameters are as follows:

[0013]

[0014]

[0015] in, This represents the total number of levels. At maximum transmission power, For minimum transmission power, For power growth step size, Indicates the first The transmit power value for each identification level, It is a set of power gradients.

[0016] In one possible implementation, at each transmission power level, the physical conditions for activating livestock tags that are within the coverage area of ​​the current power level and are not in a silent state must satisfy:

[0017]

[0018]

[0019] in, The signal power received by the tag antenna; For the first The transmit power value for each identification level; and These are the reader transmitting antenna gain and the tag receiving antenna gain, respectively. This is the minimum wake-up sensitivity threshold for the RFID chip; Let λ be the free-space path loss at a distance of λ, and λ be the wavelength. Additional degradation factors unique to the livestock and poultry environment, To identify distance.

[0020] In one possible implementation, estimating the number of remaining tags in the current power layer based on the number of successful time slots and the number of collision time slots includes:

[0021] Based on the number of successful time slots and the number of collision time slots, a modified Schoute estimation algorithm is used to calculate the instantaneous estimate of the remaining tag count. The calculation formula is as follows:

[0022]

[0023] in, This is an instantaneous estimate of the number of remaining tags in the current identified frame. This represents the number of successful time slots in the current identification frame. This represents the number of collision time slots. The collision backtracking coefficient;

[0024] The instantaneous estimate is smoothed by a first-order inertial filter to obtain the estimated number of remaining tags in the current power layer.

[0025] In one possible implementation, a first-order inertial filter is used to smooth the instantaneous estimate, and the formula for calculating the estimated number of remaining tags in the current power layer is as follows:

[0026]

[0027] in, Forgetting factor, This is an estimate of the number of remaining tags in the current power layer.

[0028] This is an estimate of the number of remaining tags in the previous identification frame.

[0029] In one possible implementation, the initial frame length is calculated based on the estimated number of remaining tags, and the parameters of the next identification frame are adaptively corrected based on a fuzzy adjustment mechanism, including:

[0030] The channel throughput and channel collision rate of the current frame are calculated based on the time-series state vector;

[0031] The channel throughput and channel collision rate are mapped to fuzzy linguistic variables. For channel throughput, the fuzzy set is defined as follows:

[0032] For the channel collision rate, the fuzzy set is defined as:

[0033] in, For the fuzzy set of channel throughput, , , These represent fuzzy linguistic variables indicating low, medium, and high channel throughput, respectively. For the fuzzy set of channel collision rates, , These are the fuzzy language vectors when the channel collision rate is low and high, respectively.

[0034] After mapping using triangles or gradient membership functions, fuzzy inference is performed based on a preset fuzzy rule base to obtain the adjustment step size of the frame length exponent.

[0035] The centroid method is used to convert the adjustment step size of the fuzzy output into a precise value, and the frame length index of the next recognition frame is updated.

[0036] The actual number of physical time slots for the next identification frame is calculated based on the updated frame length index, using the following formula:

[0037]

[0038] in, The actual number of physical time slots for the next identification frame. The frame length exponent parameter to be used for the next identification frame.

[0039] In one possible implementation, the formula for calculating the channel throughput and channel collision rate of the current frame based on the time-series state vector is as follows:

[0040]

[0041] in, For channel throughput, For channel collision rate, Number of idle time slots This represents the number of successful time slots in the current identification frame. This represents the number of collision time slots.

[0042] In one possible implementation, the centroid method is used to convert the adjustment step size of the fuzzy output into a precise value, and the calculation formulas for updating the frame length exponent of the next recognition frame are as follows:

[0043]

[0044]

[0045] =

[0046] in, The precise Q-value adjustment amount for the output of the fuzzy controller. For fuzzy output, continuous variables in the universe of discourse, Let be the membership function of the total fuzzy output set. The frame length exponent parameter used for the current identification frame. The frame length exponent parameter to be used in the next identification frame. This is an estimate of the number of remaining tags in the current power layer. This is the rounding function.

[0047] One possible implementation also includes locking the session state of successfully identified tags. When any livestock RFID tag is successfully identified, the RFID reader switches or updates the session state identifier of the successfully identified tag, causing the successfully identified tag to enter a logical silent state. In the subsequent power level identification process, the RFID reader only identifies RFID tags in the preset session state and blocks any further responses from already identified tags.

[0048] Compared with the prior art, the beneficial effects of this application are as follows:

[0049] This application achieves hierarchical identification of high-density tag groups by hierarchically controlling the reader's transmission power. It also dynamically adjusts frame time slot parameters based on collision statistics during the identification process and adaptively optimizes the time slot allocation process through a fuzzy adjustment mechanism, thereby reducing the probability of identification collisions and improving the efficiency of multi-tag identification. This application effectively reduces the channel collision probability during the identification of high-density livestock and poultry RFID tags, ensuring the integrity of individual livestock and poultry identification while shortening system inventory time, providing reliable technical support for accurate data collection and production management in smart farming scenarios. Attached Figure Description

[0050] Figure 1 A flowchart illustrating an RFID multi-tag anti-collision identification method for high-density livestock and poultry farming environments, provided as an embodiment of this application;

[0051] Figure 2 A schematic diagram of the on-site layout for an inventory check scenario applied to a pig slaughtering assembly line, provided as an embodiment of this application;

[0052] Figure 3 A schematic diagram of physical layering and tag activation state based on power gradient provided for embodiments of this application;

[0053] Figure 4 A fuzzy logic adjustment block diagram provided for embodiments of this application. Detailed Implementation

[0054] The present solution will now be described in conjunction with the accompanying drawings and specific embodiments.

[0055] Figure 1A flowchart illustrating an RFID multi-tag anti-collision identification method for high-density livestock and poultry farming environments, provided as an embodiment of this application, is shown below. Figure 1 and Figure 2 This embodiment provides an RFID multi-tag anti-collision identification method for high-density livestock and poultry farming environments, comprising:

[0056] S101, calculate the total number of power levels and construct a power gradient set according to the set power scanning parameters.

[0057] In this embodiment, for a high-density set of livestock and poultry tags, this method does not use a single full-power scan, but instead establishes a discrete gradient set of reader transmission power, which physically divides the radio frequency coverage space into K concentric, non-overlapping or partially overlapping virtual identification levels.

[0058] When performing the k-th level identification task, the reader uses power... Continuous wave is emitted. At this time, only those within the current power radiation radius... And the received signal strength Greater than the chip activation threshold A subset of tags is activated. To quantify the physical constraints between power and recognition distance, this embodiment establishes a tag activation model based on the Friesian transmission equation. In the complex electromagnetic environment of livestock farming, the physical conditions for tag activation must satisfy the following inequality:

[0059]

[0060] in: The signal power received by the tag antenna; For the first The transmit power value for each identification level; and These are the reader transmitting antenna gain and the tag receiving antenna gain, respectively. This is the minimum wake-up sensitivity threshold for the RFID chip; For the free space path loss at the distance, satisfy λ is the wavelength; This is an additional fading factor specific to the livestock and poultry environment, used to characterize the additional signal loss caused by absorption by livestock and poultry body fluids and multipath reflection from metal fences. The equations above show that... , , and environmental factors Under relatively stable conditions, the identification distance d and the transmission power They are positively correlated. Therefore, this embodiment adjusts the linear increment... It can precisely control the effective coverage radius of the radio frequency field. This will result in the total tag set. Decomposed into multiple subsets orthogonally or quasi-orthogonally in spatial dimensions:

[0061]

[0062] in, This physical model-based spatial segmentation significantly reduces the number of tags competing in each identification cycle, thus suppressing large-scale conflicts at the physical level.

[0063] Before implementing this method, an UHF RFID system conforming to the EPC C1G2 / ISO 18000-6C standard must be deployed. Reader: A high-performance four-channel reader based on the Impinj R2000 RF chip is used, fixedly installed above or to the side of the channel. Transmit antenna gain. The tag antenna gain is 9 dBi. The sensitivity threshold of the tag chip is 2 dBi. It is approximately -18dBm.

[0064] Scenario: A 2.5-meter-wide aisle is used to quickly pass through pigs wearing RFID ear tags. The tags are densely distributed, and due to the movement of the pigs and reflections from the metal fence, the distance d between the tag and the antenna and the signal fading factor are significant. Rapid and real-time changes.

[0065] After the reader is powered on, it first performs system initialization. The power scan parameters are set: minimum transmit power. 21dBm, maximum transmit power 32dBm, power increment step It is 3dBm.

[0066] The formulas for calculating the total number of power levels and constructing the power gradient set based on the set power scanning parameters are as follows:

[0067]

[0068]

[0069] in, This represents the total number of levels. At maximum transmission power, For minimum transmission power, For power growth step size, Indicates the first The transmit power value for each identification level, It is a set of power gradients.

[0070] In this embodiment, the total number of levels is 5, and the power gradient set is as follows: .

[0071] S102 controls the RFID reader to transmit power at multiple levels to the high-density distributed group of livestock and poultry RFID tags according to the power gradient in the power gradient set.

[0072] See Figure 3 In this embodiment, when the transmission power is At that time, its effective coverage radius Minimum, only the nearest subset of tags can be activated. As k increases, the coverage area expands outward in concentric circles, activating subsets of tags sequentially. This step achieves the initial splitting of the high-density tag group in physical space.

[0073] S103, at each level of transmission power, activate the livestock and poultry tags that are within the coverage area of ​​the current power level and have not been set to a silent state, form a subset of tags to be identified at the current power level, and initiate an identification frame within the current power level. Count the number of idle time slots, successful time slots and collision time slots in the current identification frame to form a time slot state vector.

[0074] In this embodiment, a channel feature extraction mechanism based on time slot state vectors is constructed. At each power level... During the identification process, the reader sends a Query command to initiate a round of identification frames. This identification frame consists of L discrete time slots. The reader demodulates the received baseband signal and statistically analyzes the time slot state vector of the current frame. :

[0075]

[0076] This vector serves as the data hub connecting the physical channel state and the upper-layer control algorithm. The physical meaning and function of each component are as follows: E represents the number of idle time slots, indicating that no tag responded in this time slot and the received power is below the noise threshold. It characterizes the degree of waste of channel resources; when the proportion is too high, it prompts the system to shorten the frame length. S represents the number of successful time slots, indicating that only one tag responded and successfully decoded the CRC checksum in this time slot. It serves as the counting benchmark for identified tags and participates in the throughput calculation. C represents the number of collision time slots, indicating that multiple tags responded simultaneously in this time slot, leading to decoding failure. By constructing... The system transforms the characteristics of analog radio frequency signals into digital statistical characteristics, providing accurate data input for subsequent Bayesian estimation and fuzzy decision-making.

[0077] S104, Estimate the number of remaining tags in the current power layer based on the number of successful time slots and the number of collision time slots.

[0078] In this embodiment, a Bayesian estimation method for the number of labels based on a modified Schoute model is proposed. To accurately calculate the time slot resources required for the next frame, this embodiment introduces a posterior probability estimation model based on collision backtracking. First, based on the acquired time slot state statistical vector... Define the estimated number of remaining tags in the current power layer. The calculation formula is:

[0079]

[0080] Where μ is the collision backtracking coefficient, used to characterize the number of potential unidentified tags behind a collision slot. To determine the baseline value of the coefficient μ, this embodiment first derives it based on the classic Dynamic Frame Slotted ALOHA (DFSA) theory. Assume the total number of tags in the reader's radio frequency field is n, and the number of slots in the current frame is L. Under the random access mechanism, the probability of any tag selecting the j-th slot is... According to the binomial distribution principle, the probability that exactly r tags respond in the j-th time slot is:

[0081]

[0082] When the number of tags Larger and longer frame When the system is in optimal throughput mode, it is in a state of adaptation. The binomial distribution converges to the expected value. Poisson distribution

[0083]

[0084] Based on the above distribution, the theoretical probabilities of the three time slot states can be calculated:

[0085] Idle probability :

[0086] Success probability :

[0087] Collision probability :

[0088] At this point, solving for μ is essentially solving for the average number of tags contained in that time slot, given that a collision has occurred. This is the conditional expectation. According to the formula for the total expectation, the derivation of this value is as follows:

[0089]

[0090] The numerator represents the expected total number of tags in all collision slots, which can be expressed as the total expected number minus the number of tags in the successful slots:

[0091]

[0092] Substituting λ=1 into the above formula, we obtain the calculated theoretical value:

[0093]

[0094] Therefore, under the ideal channel model, the theoretical coefficient of the Schoute estimation algorithm is 2.39.

[0095] It must be pointed out that the above derivation is based on the ideal assumption that all collisions can be detected and all tag signal strengths are consistent. In actual high-density livestock and poultry farming environments, there are two physical phenomena that deviate from the theory: the capture effect and multipath cancellation and absorption. This means that in livestock and poultry farming scenarios, behind each detected collision time slot, there are actually more tags to be identified than the theoretical value of 2.39. To compensate for this statistical bias caused by the unstructured environment, this invention has engineered a correction to μ, setting its value range as follows:

[0096]

[0097] Furthermore, to smooth out the jitter estimation, this embodiment uses a moving weighted average method to update the final estimate, calculated as follows:

[0098]

[0099] Where α is the forgetting factor, used to balance the weights of historical estimates and current observations. This is an estimate of the remaining number of tags in the previous identification frame. Through the above-described corrected model, the system can more accurately reconstruct the true number of tags in a high-density environment, thereby guiding subsequent time slot allocation.

[0100] S105, calculate the initial frame length based on the estimated number of remaining tags, and adaptively correct the parameters of the next identification frame based on the fuzzy adjustment mechanism.

[0101] See Figure 4 In this embodiment, the channel throughput and channel collision rate of the current frame are calculated based on the time-series state vector. The calculation formula is as follows:

[0102]

[0103] in, For channel throughput, For channel collision rate, Number of idle time slots This represents the number of successful time slots in the current identification frame. This represents the number of collision time slots.

[0104] Channel throughput and channel collision rate are mapped to fuzzy linguistic variables. For channel throughput, the fuzzy set is defined as follows:

[0105] For the channel collision rate, the fuzzy set is defined as:

[0106] in, For the fuzzy set of channel throughput, , , These represent fuzzy linguistic variables indicating low, medium, and high channel throughput, respectively. For the fuzzy set of channel collision rates, , These are the fuzzy language vectors for when the channel collision rate is low and high, respectively.

[0107] Mapping can be performed using triangles or gradient membership functions, for example, for... The membership function of the High state can be defined as:

[0108]

[0109] Fuzzy inference is performed based on a pre-defined fuzzy rule base to obtain the adjustment step size of the frame length index. Typical inference rules include:

[0110] Rule 1: IF ( is High) AND ( is Low) THEN ( Positive Big (PB)

[0111] Rule 2: IF ( is Low) AND ( is Low) AND ( is High) THEN ( is NegativeSmall, NS).

[0112] Rule 3: IF ( is High) THEN ( (is Zero, ZO).

[0113] The centroid method is used to convert the fuzzy output set obtained from fuzzy inference into precise adjustment step size values. :

[0114]

[0115] Finally, the formula for updating the Q-value of the next recognition frame is:

[0116]

[0117] The actual number of time slots in the next frame The calculation is as follows:

[0118]

[0119] in, This represents the precise Q-value adjustment amount output by the fuzzy controller, used to correct the current frame length deviation. For fuzzy output, continuous variables in the universe of discourse, Let be the membership function of the total fuzzy output set. This indicates the frame length exponent parameter used in the current identification frame. Initial source: For each power level... The first recognition frame, Based on the preliminary estimate of the number of tags in step two Initialization is typically performed, usually set to = If there is no prior estimate, it can be set to the protocol default value. Iteration source: For subsequent identification frames within this power level, The value is obtained from the previous calculation. This forms a closed-loop feedback control. : Indicates the frame length exponent parameter to be used in the next identification frame. : This indicates the rounding function, used to ensure that the Q value conforms to the integer specification of the EPC C1G2 protocol. : Indicates the actual number of physical time slots in the next identification frame.

[0120] Through the above iterative update mechanism, when the channel is congested... , Increase frame length It expands exponentially; when the channel is idle (ΔQ<0), The frame length is reduced, and this nonlinear adjustment mechanism ensures that the system can quickly converge to the optimal throughput state.

[0121] S106, based on the corrected frame length, continue to identify the remaining livestock and poultry tags in the current power layer until all tags in the current layer are successfully identified and set to a silent state, then switch to the next power layer to continue identification until all power layers have been identified.

[0122] Furthermore, this embodiment introduces a session state locking and silence mechanism. To ensure the physical effectiveness of hierarchical identification and prevent interference caused by tags already identified at low power levels responding again at high power levels, this embodiment utilizes the session mechanism in UHF RFID protocols such as EPC C1G2. The reader maintains a target session during the identification process. When any livestock tag is successfully identified, the reader immediately sends a session flip command, flipping the tag's inventory marker from state A to state B. According to the protocol specification, tags in state B remain silent to commands for state A for a set duration, emitting no backscatter signals. At each power level... During the identification cycle, the reader continuously sends inventory commands for the A status tag. Therefore, as the power from... Upgraded to Although the signal coverage area has been expanded, previously identified tags are automatically hidden because they are in state B. The reader only needs to process newly entered remote tags in the coverage area, thus achieving physical interference shielding.

[0123] The specific implementation process is as follows: Session selection and initialization: The reader selects any one of the four standard sessions (S0, S1, S2, S3). In this embodiment, Session S2 or S3 is preferred because it has a longer state retention time and is suitable for multi-level power scanning processes. In the initial state, the inventory flag bit of all tags is in state A by default.

[0124] Target Locking: At the beginning of each identification frame, the reader sends a Query command. This command includes a Target parameter, which is fixed to Target = A in this invention. This means that only tags with the inventory identifier bit in state A are eligible to respond to the reader and participate in the current ALOHA competition. State Flip and Silence: When a tag successfully sends an RN16 code in a certain time slot and receives an ACK confirmation command from the reader, the tag's internal protocol state machine automatically triggers a state flip: At this point, the inventory flag of the label changes to state B.

[0125] Physical interference shielding: During subsequent identification, the Query command issued by the reader remains Target=A. Since the identified tags are currently in state B and do not meet the activation conditions of the Query command, they will strictly remain RF silent, meaning they will not backscatter any signals or participate in time slot contention. The tag state will only revert from B to A after all power level scans are completed, the reader sends a Select reset command, or the session duration has expired.

[0126] By using this mechanism of reading only A, then changing to B, and B remaining silent, the identification of tags is precisely eliminated at the logical level. This ensures that the high-power level radio frequency energy is used only to activate unidentified tags further away, thus guaranteeing the effective implementation of the physical layering strategy.

[0127] This embodiment designs an adaptive dwell time-based hierarchical switching strategy. For each power level... The scanning duration is not fixed, but dynamically adjusted according to the label density within the layer. A switching decision condition is set: if M consecutive recognition frames satisfy... If the current layer identification is complete, then the current layer is considered to be finished. If the exit condition is not met after reaching the maximum frame limit, it means that the current layer density is extremely high. The system will automatically trigger a forced delay mechanism to continue performing multiple rounds of fuzzy adjustment and inventory at the current power until the congestion is relieved, so as to ensure that the high-density subset is completely read and resolutely prevent missed readings.

[0128] In this embodiment, "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, the simultaneous existence of A and B, or the existence of B alone. A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0129] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0130] The above description is merely a specific embodiment of this application. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the protection scope of this application. The protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A method for anti-collision identification of RFID multi-tags in high-density livestock and poultry farming environments, characterized in that, include: The total number of power levels is calculated and a power gradient set is constructed based on the set power scan parameters. The RFID reader is controlled to transmit power at multiple levels to a high-density distribution of livestock and poultry RFID tags according to the power gradient in the power gradient set. At each level of transmission power, activate the livestock and poultry tags that are within the coverage area of ​​the current power level and have not been set to a silent state, form a subset of tags to be identified at the current power level, and initiate an identification frame within the current power level. Count the number of idle time slots, successful time slots and collision time slots in the current identification frame to form a time slot state vector. The number of remaining tags in the current power layer is estimated based on the number of successful time slots and the number of collision time slots; The initial frame length is calculated based on the estimated number of remaining tags, and the parameters of the next identification frame are adaptively corrected based on the fuzzy adjustment mechanism. Based on the corrected frame length, continue to identify the remaining livestock and poultry tags in the current power level until all tags in the current level are successfully identified and set to a silent state. Then switch to the next power level to continue identification until all power levels have been identified.

2. The RFID multi-tag anti-collision identification method for high-density livestock and poultry farming environments according to claim 1, characterized in that, The formulas for calculating the total number of power levels and constructing the power gradient set based on the set power scanning parameters are as follows: in, This represents the total number of levels. At maximum transmission power, For minimum transmission power, For power growth step size, Indicates the first The transmit power value for each identification level, It is a set of power gradients.

3. The RFID multi-tag anti-collision identification method for high-density livestock and poultry farming environments according to claim 1, characterized in that, At each power level, the physical conditions for activating livestock tags that are within the coverage area of ​​the current power level and are not in a silent state must be met: in, The signal power received by the tag antenna; For the first The transmit power value for each identification level; and These are the reader transmitting antenna gain and the tag receiving antenna gain, respectively. This is the minimum wake-up sensitivity threshold for RFID. Let λ be the free-space path loss at a distance of λ, and λ be the wavelength. Additional degradation factors unique to the livestock and poultry environment, To identify distance.

4. The RFID multi-tag anti-collision identification method for high-density livestock and poultry farming environments according to claim 1, characterized in that, The number of remaining tags in the current power layer is estimated based on the number of successful time slots and the number of collision time slots, including: Based on the number of successful time slots and the number of collision time slots, a modified Schoute estimation algorithm is used to calculate the instantaneous estimate of the remaining tag count. The calculation formula is as follows: in, This is an instantaneous estimate of the number of remaining tags in the current identified frame. This represents the number of successful time slots in the current identification frame. This represents the number of collision time slots. The collision backtracking coefficient; The instantaneous estimate is smoothed by a first-order inertial filter to obtain the estimated number of remaining tags in the current power layer.

5. The RFID multi-tag anti-collision identification method for high-density livestock and poultry farming environments according to claim 4, characterized in that, The instantaneous estimate is smoothed by applying a first-order inertial filter, and the formula for calculating the estimated number of remaining tags in the current power layer is as follows: in, Forgetting factor, This is an estimate of the number of remaining tags in the current power layer. This is an estimate of the number of remaining tags in the previous identification frame.

6. The RFID multi-tag anti-collision identification method for high-density livestock and poultry farming environments according to claim 1, characterized in that, The initial frame length is calculated based on the estimated number of remaining tags, and the parameters of the next recognition frame are adaptively corrected based on a fuzzy adjustment mechanism, including: The channel throughput and channel collision rate of the current frame are calculated based on the time-series state vector; The channel throughput and channel collision rate are mapped to fuzzy linguistic variables. For channel throughput, the fuzzy set is defined as follows: For the channel collision rate, the fuzzy set is defined as: in, For the fuzzy set of channel throughput, , , These represent fuzzy linguistic variables indicating low, medium, and high channel throughput, respectively. For the fuzzy set of channel collision rates, , These are the fuzzy language vectors when the channel collision rate is low and high, respectively. After mapping using triangles or gradient membership functions, fuzzy inference is performed based on a preset fuzzy rule base to obtain the adjustment step size of the frame length exponent. The centroid method is used to convert the adjustment step size of the fuzzy output into a precise value, and the frame length index of the next recognition frame is updated. The actual number of physical time slots for the next identification frame is calculated based on the updated frame length exponent, using the following formula: in, The actual number of physical time slots for the next identification frame. The frame length exponent parameter to be used for the next identification frame.

7. The RFID multi-tag anti-collision identification method for high-density livestock and poultry farming environments according to claim 6, characterized in that, The formulas for calculating the channel throughput and channel collision rate of the current frame based on the time-series state vector are as follows: in, For channel throughput, For channel collision rate, Number of idle time slots This represents the number of successful time slots in the current identification frame. This represents the number of collision time slots.

8. The RFID multi-tag anti-collision identification method for high-density livestock and poultry farming environments according to claim 6, characterized in that, The formulas for converting the adjustment step size of the fuzzy output into a precise value using the centroid method and updating the frame length exponent of the next recognition frame are as follows: = in, The precise Q-value adjustment amount for the output of the fuzzy controller. For fuzzy output, continuous variables in the universe of discourse, The membership function for the total fuzzy output set; The frame length exponent parameter used for the current identification frame. The frame length exponent parameter to be used in the next identification frame. This is an estimate of the number of remaining tags in the current power layer. This is the rounding function.

9. The RFID multi-tag anti-collision identification method for high-density livestock and poultry farming environments according to claim 1, characterized in that, It also includes locking the session state of successfully identified tags. When any livestock RFID tag is successfully identified, the RFID reader switches or updates the session state identifier of the successfully identified tag, so that the successfully identified tag enters a logical silent state. In the subsequent power level identification process, the RFID reader only identifies RFID tags in the preset session state and blocks the response of the already identified tag.

Images